Journal of Food Engineering 101 (2010) 214–222

Contents lists available at ScienceDirect

Journal of Food Engineering

journal homepage: www.elsevier.com/locate/jfoodeng

The variable nature of Biot numbers in food drying

Sergio A. Giner a,b,c,*, R. Martín Torrez Irigoyen a, Sabrina Cicuttín a, Cecilia Fiorentini a
a
Centro de Investigación y Desarrollo en Criotecnología de Alimentos (CIDCA), UNLP-CONICET, Calle 47 y 116 (B1900 AJJ) La Plata, Argentina
b
Facultad de Ingeniería, Universidad Nacional de La Plata, Argentina
c
Comisión de Investigaciones Científicas (CIC), Provincia de Buenos Aires, Argentina

a r t i c l e i n f o a b s t r a c t

Article history: Biot numbers (Bi) relate internal and external resistances to mass or heat fluxes and are useful to
Received 23 February 2010 identify controlling mechanisms. Their variation during drying received little attention despite its
Received in revised form 25 June 2010 importance in model selection. Therefore, variable Bi were analysed in drying of wheat, a low-mois-
Accepted 4 July 2010
ture product, and in the formation of a low-calorie, dehydrated apple-leather (LCAL), a high moisture
Available online 27 July 2010
product. Wheat drying was predicted with a numerical solution from a m.c. of 0.35 kg water/kg dry
matter, to find a variable mass transfer Bi that shifted the controlling mechanism from internal–exter-
Keywords:
nal to internal. From a lower m.c. of 0.205 kg water/kg dry matter, the entire process was internally
Biot number
Drying
controlled. Drying of shrinking LCAL was solved with the energy balance. The mass transfer Bi varied
Food from 0.5 to 1600, while the heat transfer Bi, only from 0.25 to 0.5, and a modified version, from 0.25
to 0.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction
The Biot number is a dimensionless group that relates a mea-
sure of the rate of internal heat conduction or mass diffusion in sol-
ids with a measure of the rate of external convection. It appears in
partial differential equations of transport phenomena, when the
surface boundary condition relating internal and external fluxes
is expressed in dimensionless form.
A typical graph used by many chemical and food engineers was
published first by Gurney and Lurie (1923), who analytically solved
the unsteady state heat conduction equation with an infinite series,
providing the predictions of dimensionless local temperature in-
side the solids (the dependent variable), as a function of the dimen-
sionless time for several dimensionless positions (the independent
variables) and several Biot numbers. An inherent assumption is
that the Biot number remains constant during the process. The
analytical series eventually converges in one term at long times,
allowing straight lines to be plotted in a semilog graph. These
graphs were and are still used to solve heat transfer problems for
regular geometries as slab, infinite cylinder and sphere, as well
as three-dimensional regular geometries (Singh and Heldman,
1993).
The risk of assuming the validity of the heat and mass transfer
analogy for using these graphs to solve mass transfer processes as
* Corresponding author at: Centro de Investigación y Desarrollo en Criotecnología
de Alimentos (CIDCA), UNLP-CONICET, Calle 47 y 116 (B1900 AJJ) La Plata,
Argentina. Tel.: +54 221 4249287; fax: +54 221 42890741.
E-mail address: saginer@ing.unlp.edu.ar (S.A. Giner).
drying is that the mass transfer Biot number may vary during the
process (Giner and Mascheroni, 2001). Another problem that usu-
ally appears in drying is that moisture content in solids is not an
uniform field across the solid–fluid interface unlike the vapour
pressure, so a sorption relationship must be used to transform
moisture contents at the solid surface into vapour pressure (Crap-
iste and Rotstein, 1997).
Adhikari et al. (2005), during a research aimed at predicting
stickiness during spray drying of a sugar-rich solution, have pre-
dicted the evolution of Biot numbers during the process in a
120 lm initial diameter drop. They found internal control for
the mass transfer rate from the start, and a predominantly exter-
nal control for heat transfer. However, additional contributions
are required to study the definition and variation of the heat
and mass transfer Biot numbers in larger food particles in order
to assess changes in the controlling mechanisms. This feature is
in itself relevant, and, besides, provides better background for
selection of suitable mathematical models and methods of solu-
tion. This is especially important nowadays, where much of the
contemporary drying research resorts to empirical kinetic mod-
els, which do not explain the drying mechanisms (Akpinar et
al., 2003).
Therefore, the objectives of this work were (1) to calculate mass
and heat transfer Biot numbers during drying of foods and (2) to
analyse their behaviours.
To this end two case studies are presented: wheat drying
kinetics, using data from a previous study, and formation of a
low-calorie, apple-leather during hot-air dehydration, a novel
product.

0260-8774/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2010.07.005
 S.A. Giner et al. / Journal of Food Engineering 101 (2010) 214–222 215

Nomenclature

A transfer area of food in Eq. (16), m2 nt time index for numerical integration, t = (nt  1) Dt
aw water activity, decimal Nmh parameter in Eq. (3)
Bic heat transfer Biot number, dimensionless ps saturation vapour pressure of pure water, Pa
Bicd heat transfer Biot number, defined in Eq. (19), dimen- pv vapour pressure at food surface, Pa
sionless pva partial vapour pressure in the air bulk, Pa
Bim mass transfer Biot number, dimensionless pvad dimensionless vapour pressure at food surface
Cmh parameter in Eq. (3) pv0 initial vapour pressure at food surface, Pa
Cp specific heat of food defined in Eq. (17), J kg1 dry mat- r radial coordinate in a wheat grain, m
ter K1 R equivalent spherical radius in wheat, m
cw local concentration of water in food in Eq. (13), kg m3 rad dimensionless radial coordinate in a wheat grain
cw0 initial local concentration of water in food, kg m3 Rg gas constant, J mol1 K1
Deff effective diffusion coefficient of water in the food, t time, s
m2 s1 T grain temperature in Eq. (3), °C
Deff0 effective diffusion coefficient of water in food at W0 de- Ta air temperature, °C
fined in Eq. (20), m2 s1 Tm time-dependent average temperature of food in Eq. (16),
D1 Arrhenius preexponential factor in Eq. (22), m2 s1 °C
D11 parameter in Eq. (4) Tm0 initial food temperature, °C
D12 parameter in Eq. (4) x coordinate along food thickness in Eq. (13), m
Ea activation energy in food drying, J mol1 W local moisture content of food, kg water/kg dry matter
hT air-food heat transfer coefficient, W m2 K1 Wad dimensionless moisture content of food
hTd air-food heat transfer coefficient, defined in Eq. (19), We equilibrium moisture content, kg water/kg dry matter
W m2 K1 W0 initial moisture content of food, kg water/kg dry matter
i space index for numerical integration of Eq. (13), x = (i
 1) Dx Greek symbols
I total number of points along x for numerical integration DHws heat of desorption of water from food in Eq. (16), J kg1
of Eq. (13) Dx distance step for numerical integration of Eq. (13), m
Kmh parameter in Eq. (3) Dt time step for numerical integration of Eq. (13), s
kp air-food mass transfer coefficient, q density of food at moisture content W, kg m3
kg vapour m2 s1 Pa1 qd density of food dry matter in Eq. (19), kg m3
kT food thermal conductivity, W m1 K1 qd0 ratio of the dry matter in the food to its initial volume,
L variable thickness of food during drying, m kg m3
L0 initial food thickness in Eq. (19), m q0 initial density of food, kg m3
ms food dry mass in Eq. (16), kg qw density of liquid water, kg m3
mwev rate of water evaporation in food in Eq. (16), kg s1
n parameter in Eq. (20)

2. Mathematical modelling On the other hand, water activity aw = pv/ps and W, the local
moisture content in the grain (kg water/kg dry matter), are related
2.1. Model describing wheat drying kinetics

By assuming the grain as a sphere with an average diffusion

coefficient, Deff, the water transport inside the grain may be repre-
sented by the combination of the microscopic mass balance and
the first Fick’s law of diffusion (Crapiste and Rotstein, 1997):
!
@W @ 2 W 2 @W
¼ Deff þ ð1Þ
@t @r2 r @r

With the following initial and boundary conditions.

t¼0 W ¼ W0 0rR ð2:1Þ

@W
r¼0 ¼0 t>0 ð2:2Þ
@r

@W 
r¼R  qd0 Deff ¼ kp ðpv  pva Þ t>0 ð2:3Þ
@r r¼R

where qd0 is the ratio of the dry mass to the grain volume, qd0 = q0/
(1 + W0), q0 being the grain particle density at the initial moisture
content, W0. Wheat density data measured by Giner and Denisienia
(1996) was analysed here to develop a third-grade polynomial cor-
relation with moisture content (Fig. 1).
Fig. 1. Experimental (d) and predicted (—) data of wheat grain density (q) as a
function of moisture content (W) in kg water/kg dry matter. The predictive
expression used was an empirical third-grade polynomial, q = 23463.5
W3  11872.3 W2 + 1370.8 W + 1310.0. The coefficient of determination, r2 was of
0.999. The correlation is valid for moisture contents W between 0 and 0.3 kg water/
kg dry matter.
216 S.A. Giner et al. / Journal of Food Engineering 101 (2010) 214–222

by the sorption isotherm. Symbol pv is the water vapour pressure The formal definition of the mass transfer Biot becomes,
at the food surface, while ps is the saturation vapour pressure of
k R
pure water, predicted with a correlation published earlier (Giner Bim ¼ h p i ð9Þ
1
and Mascheroni, 2001). The symbol pva is the partial pressure of ps
@W
@aw
qd0 Deff
r¼R
water vapour in the air bulk and kp is the mass transfer coefficient
in kg water vapour/m2 s Pa. The modified Henderson equation was This dimensionless number expresses the ratio of the internal
selected for the aw–W relationship, which has been widely used to resistance to mass transfer to the external one. The internal is rep-
predict drying processes in grains. Considering that aw = pv/ps. resented by the grain equivalent radius (R) and the reciprocal of
2   31=Nmh the diffusion coefficient (Deff) while the external, by the reciprocal
Ln 1  ppv of the mass transfer coefficient (kp).
W ¼ 0:014 5
s
ð3Þ By using Eq. (3), the following derivative is found:
K mh ðT þ C mh Þ  
     1  
For breadmaking wheat, Kmh, Cmh and Nmh are 2.301  105, @W 1 1 N mh 1
¼ 0:01 1 ½Lnð1  aw Þ 
55.815 and 2.2857, respectively (Brooker et al., 1992). Both in @aw N mh K mh ðT þ C mh Þ 1  aw
wheat (Giner and Mascheroni, 2001) and soybean drying (Gely ð10Þ
and Giner, 2007), no quantitative differences were observed in By replacing the expressions of Eq. (7)–(9) into Eq. (6), the def-
the predicted mean grain moisture content as a function of time, inition of the mass transfer Biot number, Bim becomes.
between the isothermal model of Eqs. (1)–(3) and the non-isother-
kp R
mal model that adds heat transfer coupling. The enormous differ- Bim ¼ 2   3
   1  
ences between heat and mass transfer rate, the small size of 14 N mh
ps
0:01 N1mh  1 K mh ðTþC
1
mh Þ
½Lnð1  aw Þ  1aw 5
1
qd0 Deff
grains and its low-moisture nature were found to be the causes
r¼R
of this nominally isothermal drying. Therefore, only the mass
ð11Þ
transfer model was solved for wheat. All properties used in wheat
were taken from Giner and Mascheroni (2001), except the diffusion In a previous work, Parti (1993) had defined Bim as ky R/(qDeff),
coefficient which was developed by the same authors in a subse- where ky is the mass transfer coefficient for a flux equation that
quent article (Giner and Mascheroni, 2002). utilises the driving force in air humidity units. While this variable
  is directly related with the partial pressure of vapour in air, the
Ea
Deff ¼ ½D11 þ D12 ðW 0  0:1891Þ exp  ð4Þ expression of Parti does not include the conversion between food
Rg ðT a þ 273:16Þ
moisture content and air humidity or vapour pressure at the inter-
where D11 = 5.046  107 m2 s1 and D12 = 54.44  107 m2 s1/ face and is therefore incorrect.
(kg water/kg dry matter). The value of the activation energy was On the other hand, if Biot numbers are predicted for non-iso-
Ea = 27,184 J mol1. The symbol Ta is the drying air temperature, thermal systems and/or using moisture-dependent diffusion coef-
in °C. ficients, the value of Deff must be evaluated at the temperature
and/or local moisture content prevailing at the surface at any time.
2.1.1. Calculation of the mass transfer Biot number Besides, models predicting shrinkage during drying, must update
To establish an uniform field variable across the grain–air inter- the product characteristic dimension in each time step of the
face, the surface boundary condition Eq. (2.3) was expressed in an numerical calculation. Therefore, apart from the variable nature
equivalent, alternative form, by including the following identity. of the conversion factor (oW/oaw), possible variations in Deff and
   R mean that the mass transfer Biot number would change consid-
@W @W @pv
 ð5Þ erably during drying.
@r @pv @r
which, once included into the surface boundary condition (Eq. 2.1.2. Conventional heat transfer Biot number
(2.3.), becomes The heat transfer Biot number.
!
@pvs kp hT R
r¼R  qd0 Deff ¼ ðpv  pva Þ t>0 ð6Þ Bic ¼ ð12Þ
@r @W kT
@pv

where qd0 is the dry matter divided by the grain volume. To find the was calculated at the same conditions as for the mass transfer coun-
mass transfer Biot number, variables must be expressed in dimen- terpart. As hT, the interfacial heat transfer coefficient (Giner and
sionless form Mascheroni, 2001) have a weak temperature dependence, and as
kT is a smooth function of moisture content (Sokhansanj and Bruce,
W  We r pv  pva 1987), Bic is not expected to vary considerably during drying.
W ad ¼ ; r ad ¼ ; pvad ¼ ð7Þ
W0  We R pv0  pva
where R, is the grain equivalent radius, was taken as 2.0  103 and 2.2. Drying model for the formation of a low-calorie apple-leather
1.8  103 m for initial moisture contents (W0) of 0.35 and 0.205 kg (LCAL)
water/kg dry matter, respectively (Giner and Mascheroni, 2002).
The symbol We stands for the equilibrium moisture (kg water/kg 2.2.1. Product characteristics
dry matter), calculated with Eq. (3) for the temperature and relative The LCAL was formulated to experience gelation during drying
humidity of the drying air. Knowing that pv is the water vapour by the saccharide-acid-high methoxyl pectin mechanism. A previ-
pressure at the solid surface (r = R), pv0 represents its initial value. ous formulation (Leiva Díaz et al., 2009) composed of peeled apple
Vapour pressure drops from pv to pva in the air film surrounding puree, sucrose and citric acid, was modified by replacing sucrose
the grain. The derivative of moisture with respect to vapour by polydextrose which is a polysaccharide and thickener that pro-
pressure can be more easily evaluated from the sorption isotherm vides only a quarter of the energy intake of sucrose (Cicuttin and
model if we consider the following identity Giner, 2006). The initial formulation had the following composition
    (% w/w): water, 80 (or 4 kg water/kg dry mater); carbohydrates,
@W 1 @W 8.86; proteins, 0.27; lipids, 0.087; citric acid, 0.17; polydextrose,
 ð8Þ
@pv ps @aw 10; sucralose, 0.02. It was dried in a forced convection oven with
 S.A. Giner et al. / Journal of Food Engineering 101 (2010) 214–222 217

dT m
ms C p ¼ hT AðT a  T m Þ  DHws mwev ð16Þ
dt
where Tm is the average product temperature and °C, Ta the drying
air temperature. The value of DHws, the heat of desorption of water
from the food, was considered the same as that for pure water in the
moisture range considered, as observed in sugar-rich products by
Kaymak-Ertekin and Gedik (2004). In turn, Cp is the variable specific
heat based on the product dry mass, (J kg1 dry matter °C), whose
expression was developed by rearranging the well-known Siebel
correlation for high moisture foods (Mohsenin, 1980)

C p ¼ 837:4 þ 4187 W ð17Þ

The symbol mwev is the rate of water evaporation i.e., kg water
Fig. 2. Photograph of the dehydrated low-calorie, apple pectic gel, with a moisture vapour/s, which for using in Eq. (16) can be calculated by the inter-
content of 0.50 kg water/kg dry matter and a water activity of 0.86. When removed, facial mass transfer equation:
it occupied the whole base of a 0.13  0.18 m tray. Product thickness was of about
2.5 mm. mwev ¼ kp Aðpv  pva Þ ð18Þ
The initial condition to solve Eq. (16) was t = 0; Tm = Tm0, the ini-
tial product temperature.
an air velocity of 0.2 m/s from its initial moisture content up to a
final value of 0.5 kg water/kg dry matter, corresponding to a water
2.2.4. Product properties
activity of 0.86, which prevents growth of pathogenic bacteria
2.2.4.1. Shrinkage relationship. Volume of product decreases linearly
(Beuchat, 1981, 1983). A photograph of the product, is given in
with the decrease in moisture content, in agreement with previous
Fig. 2. It must be indicated that the LCAL provides 667.1 kJ/100 g
reports (Leiva Díaz et al., 2009). Shrinkage was one-dimensional,
of finished product (Cicuttin and Giner, 2006), 42% less than the su-
being the thickness the only shrinking dimension in agreement with
crose-added apple-leather developed by Leiva Díaz et al. (2009).
observations for gels by Kechaou et al. (1987). Therefore, the volu-
metric ratio V/V0 becomes the thickness ratio L/L0. Shrinkage was
2.2.2. Diffusion equation
considered equivalent to the volume of water evaporated, so the fol-
The diffusion equation with variable diffusion coefficient, con-
lowing relationship was utilised for this LCAL, based on a similar
sidering product shrinkage involves a coupled model of heat and
model previously developed for sucrose-added, apple-leather (Leiva
mass transfer in a plane sheet, formed by the microscopic mass
Díaz et al., 2009).
balance considering product shrinkage plus the macroscopic
energy balance. This was solved by a combination of the explicit V L W0 1
finite-difference and the fourth-order Runge Kutta methods, ¼ ¼1  þ  W ð19Þ
V 0 L0 1 W0
qw q þ q qw q þ Wq 0
1
respectively. The diffusion equation is as follows: d w d w

 2 where W0 = 4 kg water/kg dry matter. The value of the dry matter

@cw @ 2 cw @Deff @cw
¼ Deff þ ð13Þ density, qd was considered equal to that for sucrose-added, apple
@t @x2 @cw @x
leathers, i.e., 1382 kg m3. The symbol qw stands for the density
The initial and boundary conditions, are: of liquid water.

t ¼ 0 cw ¼ cw0 06x6L ð14:1Þ

2.2.4.2. Diffusion coefficient. This transport parameter, Deff, was cor-
related with the volumetric water concentration by Cicuttin and
@cw
x¼0 ¼0 t>0 ð14:2Þ Giner (2006). They have observed that analytical solutions for dif-
@x
fusion for plane sheets (Crank, 1975; Leiva Díaz et al., 2009), con-
@cw sidering constant volume and diffusion coefficient, predicts the
x¼L  Deff ¼ kp  ðpv  pva Þ t > 0 ð14:3Þ drying curve of apple leathers reasonably well. Analytical solutions
@x
include the dimensionless time or Fourier number Deff t/L2 and, as
where L is the variable thickness of the LCAL, Deff is the diffusion gels evidently shrink, it was considered that the ratio of Deff to L2
coefficient of water in the forming gel (m2 s1), dependent on the must remain reasonably constant on an average basis so Deff must
local water concentration, cw in kg water/m3 product, and tempera- also decrease during drying. To extend this concept to a local basis,
ture, T in °C. The symbol x stands for the coordinate in the thickness it can be proposed that the ratio of the local diffusion coefficient,
direction (m). The relationship between local water concentration Deff to the instantaneous total thickness L is such that Deff/Ln is a
and local moisture content in kg water/kg dry matter is: constant. Then, the relationship used to determine n was as
follows.
cw
W¼ ð15Þ
q  cw Deff0 Deff
¼ n ð20Þ
Ln0 L
where q stands for product density.
Using kinetic data at short times, where shrinkage was ne-
2.2.3. Macroscopic energy balance glected, the initial diffusion coefficients Deff0 were fitted with an
Preliminary numerical solutions using a microscopic heat bal- analytical solution for Bi = 2 (Leiva Díaz et al., 2009). These coeffi-
ance coupled with the mass balance showed that the temperature cients were considered only a function of temperature obeying the
profiles are almost uniform, so that the problem of describing Arrhenius law. Then, data at long times were predicted with an
product temperature as a function of time can be tackled with a analytical series for internal control. Using the final moisture con-
macroscopic energy balance. tent and sheet thickness, L, the final diffusion coefficient Deff was
218 S.A. Giner et al. / Journal of Food Engineering 101 (2010) 214–222
estimated. Then the power n was calculated by Eq. (20) to be 1.58.
Therefore, the diffusion coefficient relationship was now the
following.
 1:58
L
Deff ¼ Deff0
L0
where Deff0, as indicated earlier, was an Arrhenius function of tem-
perature, only. Using Eq. (19) in Eq. (21) and rearranging, the final
expression of the local diffusion coefficient in the presence of
shrinkage was obtained.
  
Ea
Deff ¼ D1 exp
Rg ðT m þ 273:16Þ
( )1:58
0:28067
 ð22Þ
1  0:71933  cwqð1þW
W0
0Þ
0
where D1 = 7.547  107 m2 s1 is the preexponential factor, while
Ea, the activation energy, was found to be 19,715 J mol1. The factor
enclosed by square brackets is Deff0, while the multiplier written in
curly brackets, represents the ratio of L/L0. The effective diffusion
coefficient increases for increasing water concentration cw and
product temperature, Tm. The correlation of Eq. (22) was developed
by Cicuttin and Giner (2006) and follows structural collapse con-
cepts of a products in the rubbery state (Aguilera, 2005). Symbol
q0 stands for the initial product density = 1058.5 kg m3. To convert
from moisture content on dry matter basis to water concentration
in kg m3, Eq. (15) was used in the form
 
W
cw ¼ q
1þW
Concerning the dependence of total product density with mois-
ture content, the model developed earlier by Leiva Díaz et al.
(2009) was utilised, which is related with the shrinkage model of
Eq. (19). The density model predicts:
q d ð1 þ W Þ
q¼
1 þ qqd W
w interval was divided in two, leading to high and low-moisture Hal-
2.2.4.3. Air-product heat and mass transfer coefficients. These were
measured in previous experiments (Cicuttin and Giner, 2006) of
steady-state water evaporation carried out in the same
mechanical-convection oven at an air velocity of 0.2 m/s .The
average heat transfer and mass transfer coefficient were found to
be hT = 21.9 W m2 K1 and kp = 1.51  107 kg water vapour/
(m2 s Pa). Existing correlations differ considerably in the low air
velocity zone (Welti-Chanes et al., 2003; Marinos-Kouris and Mar-
oulis, 1995), thus the need of the experimental determinations.
2.2.4.4. Variable thickness. The domain of Eq. (14), i.e., x from 0 to L
was discretised in I nodes, so I  1 distance intervals Dx were cre-
ated, with x = (i  1) Dx. Then, the space index ‘‘i” varies from 1 to I
for nodes and from 1 to I  1 for intervals. The time, in turn, was
discretised as t = (nt  1) Dt. Therefore the leather thickness at
any time during drying was calculated by:
X
I1
L¼ Dxi;nt
i¼1
Each space interval is reduced after each time interval using the
shrinkage expression (Eq. (19)).
2.2.5. Calculation of the heat transfer Biot number during drying
Given the absence of published data for this type of leather,
the thermal conductivity for apples in J m1 K1 was utilised
(Mohsenin, 1980).
W
k ¼ 0:148 þ 0:493 ð26Þ
1þW
Therefore, the expression used to calculate the time-varying heat
transfer Biot number is
ð21Þ P
hT L hT I1
i¼1 Dxi;nt
Bic ¼ ¼ W
ð27Þ
k 0:148 þ 0:493 I;nt 1þW I;nt
The peak observed in Fig. 6 reflects the interplay occurring in Eq.
(26) as a function of time between a decreasing thickness in the
numerator and a decreasing thermal conductivity in the denomina-
tor. Both parameters diminish as a consequence of the reduction in
the surface moisture content. Nevertheless, the order of magnitude
keeps unaltered, since the values range from 0.25 at the beginning,
to 0.55 at the peak and then reduce back to almost 0.30 at the end of
the drying process.
2.2.6. Calculation of the mass transfer Biot number
By using the formal definition established in Eq. (11) and adapt-
ing it to this case study where variable diffusion coefficient, and
variable leather thickness are predicted as a function of time, a spe-
cific definition of the mass transfer Biot number for shrinking
materials was reached using Eqs. (14.3), (23) and (24), which
was computed as follows:
PI1
kp i¼1 Dxi;nt
Bim ¼ ( ) ð28Þ
 
1 @W I;nt DI;nt
ps @aw
qd q 2
ð1þqwd W Þ
ð23Þ The derivative of moisture content at the surface with respect to
water activity was calculated in this LCAL with a Halsey equation,
whose parameters were previously fitted to experimental data
measured by an Aqualab 3TE water activity meter (Cicuttin and
Giner, 2006). No net effect of temperature was found, in agreement
with Leiva Díaz et al. (2009). Moisture contents were measured in
atmospheric oven at 105 °C until constant weight (Greensmith,
ð24Þ 1998). In this high initial moisture product, the water activity
sey models (see Appendix).
3. Results and discussion
3.1. Wheat drying
3.1.1. Variation of the mass transfer Biot number for high moisture
contents
The exact analytical solution for Eq.(1), with initial and bound-
ary conditions (2.1), (2.2), (2.3) (Crank, 1975), assumes that the
Biot number is constant, i.e., not time-dependent. Few if any con-
tributions were provided by the literature to depict the variation
of the mass transfer Biot number during drying. For this purpose,
the diffusion equation mentioned above was numerically solved
by an implicit finite-difference scheme. The radius domain was
discretised in 31 nodes, using a time step of 1 s. A Matlab program
was written for this purpose, being the predictions completely
stable. The mass transfer Biot number was studied as a function
of time for wheat drying, firstly from an initial moisture content
ð25Þ of 0.35 kg water/kg dry matter. This unusually high water content
for wheat was selected in order to clarify the predicted behaviour
plotted in Fig. 3.
Being the mass transfer coefficient, sphere radius, dry matter
density relationship and effective diffusion coefficient assumed
constant, the only variable factor is the derivative of moisture con-
tent with respect to water activity (hereafter slope). This slope is
placed in the denominator of Eq. (10), so the mass transfer Biot de-
pends reciprocally on it. The behaviour of oW/oaw is typical of a
 S.A. Giner et al. / Journal of Food Engineering 101 (2010) 214–222 219

0.4

Surface moisture, kg water/kg dry matter

0.35

0.3

0.25

0.2

0.15

0.1

0.05

0
0 10 20 30 40 50 60
Tiempo, s
Fig. 4. Predicted surface moisture content as a function of time during drying, for
the conditions of Fig. 3. Surface equilibrium would be approached in some 35 s, and
strict internal control to the mass transfer rate would prevail from then on.

to commercial dryers, the behaviour of the mass transfer Biot num-

Fig. 3. Mass transfer Biot number as a function of time during drying of a wheat
seed initial moisture content: 0.35 kg water/kg dry matter. (Top) General behaviour
and (bottom) initial behaviour, magnified. The Biot number over the first seconds of
drying suggests internal–external control. Operating conditions: initial moisture
content of wheat, 0.35 kg water/kg dry matter (25.93% w/w); air temperature,
60 °C; air relative humidity, 0.08; air mass flow rate, 0.30 kg m2 s1; and grain
equivalent radius, 2.0  103 m.
ber as a function of time is shown in Fig. 5.
Though comparable with the variation observed for high mois-
ture contents in Fig. 3, the mass transfer Biot number in Fig. 5
starts with a value of about 1200, which denotes a strict internal
control, in agreement with previous literature on grain drying
(Bruce, 1985; Gastón et al., 2004). Therefore, the range of variation
in the mass transfer Biot number during drying seems to be related
with the difference between initial and equilibrium moisture con-
tents. Unlike in the results for high moisture content (Fig. 3), all the
variation shown in Fig. 5 including the peak is mechanistically
irrelevant, since it occurs entirely in the region where definite
internal control prevails. The lack of variation after about 15 s indi-
cate that equilibrium was reached at the surface very rapidly.
5000

well-founded sorption relationship for starch-based foods: it tends

to 1 for both aw ? 0 and aw ? 1. Thus, according to Eq. (10), Bim 4000
Mass Transfer Biot Number

will be very low initially, will then increase and decrease again.
However, the decrease is mechanistically unimportant, since it oc-
curs well inside the zone of strict internal control. The variation of
the derivative would explain the predicted peak in the graph of the 3000
mass transfer Biot in Fig. 3 (top) as a function of time. The initial
predicted value of this Biot number under these conditions was
2.1, suggesting that control of the mass transfer rate is initially 2000
shared between the internal and the external resistances. How-
ever, this state of affairs lasts less than 20 s (Fig. 3 bottom) as inter-
nal control takes over from then on, i.e., the mass Biot increases
1000
dramatically with time. Another evidence of the sudden increase
of the internal resistance to mass transfer within the first 30 s is
provided by Fig. 4, where the surface moisture content is predicted
to drop very sharply from the initial moisture content, towards the 0
equilibrium value with the drying air conditions. 0 10 20 30 40 50
Time, s
3.1.2. Variation of the mass transfer Biot number at lower, practical
Fig. 5. Mass transfer Biot number as a function of time during drying of a wheat
moisture contents grain. Operating conditions: initial moisture content of wheat, 0.205 kg water/kg
At an initial moisture content W0 of about 0.205 (17% w/w), dry matter (17% w/w); air temperature, 60 °C; air relative humidity, 0.08; air mass
which is a typical water content of freshly harvested wheat fed flow rate, 0.30 kg m2 s1; and grain equivalent radius, 1.8.0  103 m.
220 S.A. Giner et al. / Journal of Food Engineering 101 (2010) 214–222

1
0.9
0.8
Heat Transfer Biot Number

0.7
0.6

0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 Fig. 7. Conventional (—) and modified (—) heat transfer Biot number predicted as a
function of time during the drying of low-calorie, apple leathers. The modified
1.2 number suggests a more externally controlled heat transfer rate during drying than
that predicted by the conventional definition. This emphasizes the mass transfer-
controlled nature of food drying.
1
Heat Transfer Biot Number

3.2. Formation of a low-calorie apple-leather by hot-air dehydration

0.8
3.2.1. Analysis of the conventional and a modified heat transfer Biot
number
0.6 As the original definition of the heat transfer Biot number for
this product (Eq. (26)) is strictly valid in the absence of mass trans-
fer, it may not be suitable for drying processes. In this regard, Chen
0.4
and Peng (2005) have proposed a modified heat transfer Biot num-
ber, which can be termed ‘‘heat transfer Biot number for drying”
0.2 (Bicd).
h i
hT DHws kp ðpv pva Þ
hTd L ðT a T m Þ
L
0 Bicd ¼ ¼ ð29Þ
kT kT
0 10 20 30 40 50
Time , s where hTd is the ‘‘net” heat transfer coefficient in a drying operation,
giving its reciprocal a measure of the effective external resistance to
Fig. 6. Conventional heat transfer Biot number for wheat drying at 60 °C. Initial the heat flux entering the product. As part of the heat is employed
moisture content = 0.35 kg water/kg dry matter (top) and 0.205 (bottom). Relatively
for surface water evaporation, the net heat influx is lower than in a
small changes occur during drying, and all within the same apparent controlling
mechanism: internal–external.

8.E-10
Effective diffusion coefficient, m 2/s

7.E-10
3.1.3. Evolution of the heat transfer Biot number in wheat drying
For an initial moisture content of 0.35 kg water/kg dry matter,
6.E-10
the variation in the heat transfer Biot number (Fig. 6 top) is gov-
erned by the variation of the surface thermal conductivity. The
5.E-10
lower the surface moisture content, the lower the thermal con-
ductivity and the higher the Biot number (see Eq. (12)): in the
4.E-10
first seconds, the surface moisture content in Fig. 4 decreases
only slightly and then the heat transfer Biot number remains
3.E-10
substantially constant. However, then the surface moisture
drops sharply to reach a constant value, and, simultaneously,
2.E-10
the Biot increases suddenly to reach a value of about 0.95 after
some 30 s.
1.E-10
Concerning wheat drying from an initial moisture content of
0.205 kg water/kg dry matter, the evolution of the heat transfer
0.E+00
Biot number is simpler. From an initial value of 0.6, it increases
0 1 2 3 4 5
to 0.95, to remain then constant due to attainment of equilibrium
conditions. The final value is the same for both drying conditions, Moisture content, kg water/kg dry matter
since it depends, through the thermal conductivity, on the surface
Fig. 8. Diffusion coefficient for low-calorie apple leathers as a function of moisture
moisture content, which tends to a value in equilibrium with the content for product temperatures of 30 (. . .), 50 (—) and 70 °C (—) as predicted with
same air conditions. the correlation developed for this shrinking food.
 S.A. Giner et al. / Journal of Food Engineering 101 (2010) 214–222 221

1800 values, almost zero, denoting an external control. This is congruent

with predictions of uniform temperature profile inside the leather
1600 (though not constant with time). The increase and decrease of both
heat transfer Biot numbers along the course of drying reflects the
1400
Mass transfer Biot number

combined influences of decreasing thermal conductivity at the sur-

1200 face and decreasing thickness. As surface conductivity decreases
faster than thickness, both Biots first increase. Then, as conductiv-
1000 ity remains almost constant and thickness keeps decreasing, both
Biots tend to diminish.
800
The variable diffusion coefficient of the reduced-calorie apple-
600 leather (Eq. (22)) was plotted as a function of moisture content
for temperatures of 30, 50 and 70 °C in Fig. 8. The plots indicate
400 that both independent variables contribute considerably to the
variation of the effective diffusion coefficient during drying.
200
The evolution of the mass transfer Biot number in Fig. 9 (top)
0 shows low initial values, increasing to a situation of strong internal
0 50 100 150 200 250 300 350 400 450 500 control with time. The predicted maximum reflects the simulta-
Time, min neous effects of a decreasing diffusion coefficient and decreasing
60 factor (1 + (qd/qw)W)2 both evaluated at the surface (the factor ap-
pears as a consequence of adimensionalization of boundary condi-
50 tion (14.3)), a variable slope of the sorption relationship (oW/oaw)
and a decreasing thickness. Mechanistically, the relevant changes
Mass transfer Biot number

occur in the first 30–40 min, being the succeeding variations

40 unimportant, as they occur within the zone of strict internal con-
trol. However, in order to observe the initial behaviour, Fig. 9 (mid-
dle) shows the state of affairs for the first 35 min of drying, where
30
the Biot number changes from low values to about 50, almost in
the zone of pure internal control. Hence the duration of the zone
20 of internal–external control is about 35 min. However it would
be important to detect the nature of controlling mechanism at
the start of drying, i.e., if internal–external or external. In this re-
10
gard, Fig. 9 (bottom) indicates that the initial control to the mass
transfer rate is internal–external, starting with a mass transfer Biot
0 of 0.5 and rapidly evolving to a value of about 5. Therefore, no con-
0 5 10 15 20 25 30 35 vective, external control is predicted to prevail in any interval of
Time, min the drying of low-calorie apple leathers. This is in agreement with
5 an analysis carried out with analytical, Biot-dependent solutions,
4.5 in a previous article (Giner, 2009).
On the basis of the predictions described above, the highly
4 variable nature of the mass transfer Biot number suggests that ana-
Mass transfer Biot number

lytical solutions using constant Biot number are not a realistic

3.5 representation of the mass transfer process because, at most, it
3 can only be valid for a limited period. These analytical solutions,
however, are suitable for heat transfer problems involving neither
2.5 mass transfer nor considerable variation of product volume or
structural characteristics.
2

1.5 4. Conclusions

1 In wheat drying, variation of the Biot number during drying is

relevant at high initial moisture contents, e.g., 0.35 kg water/kg
0.5
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 dry matter. In the first minute, drying rate control shifts from inter-
Time, min nal–external to internal, keeping as such from then on.
In wheat drying at practical initial moisture contents, e.g.,
Fig. 9. Mass transfer Biot number as a function of time during drying of low-calorie, 0.205 kg water/kg dry matter, the mass Biot number is very high
apple-leather. (Top) Entire drying process; (middle), first 35 min; (bottom) first 6 s. from the beginning of the drying process, drying rate control is
The predicted Biot number at the start of drying is about 0.5.
internal throughout and variations are irrelevant.
In view of both behaviours in wheat drying, analytical solutions
situation with no drying. Therefore, from the viewpoint of the Biot may be used for low initial moisture content grain. In the case of
number analysis, the situation is equivalent to a decrease in the high moisture contents, the shifting drying rate control and the
heat transfer coefficient, and, therefore, to an increase of the exter- possibility of shrinkage suggests the use of numerical solutions.
nal resistance to heat transfer. In a high moisture product as a low-calorie apple-leather, the
The behaviours of both conventional and modified heat transfer mass transfer Biot number varies during drying in a complex man-
Biot numbers are presented in Fig. 7 for the low-calorie apple- ner, starting and keeping internal–external control of the drying
leather: while the conventional Biot keeps within a zone of rate for a period as long as 35 min at an air temperature of 60 °C.
internal–external control, the modified Biot reaches very low Mass transfer rate is internally controlled from then on. Therefore,
222 S.A. Giner et al. / Journal of Food Engineering 101 (2010) 214–222
an analytical solution of a diffusion equation with internal control
would be defective for the initial 35 min of drying and is therefore
not recommended.
The heat transfer Biot number both in wheat and low-calorie
apple-leather keeps within the internal–external control region
during the drying process.
A modified version of the heat transfer Biot for drying tends
asymptotically to zero during the process because the heat influx
to the product is lower than in pure heat transfer due to the frac-
tion of heat demanded for water evaporation at the surface.
This is congruent with the predictions of practically flat internal
temperature profile during drying of foods. The modified heat
transfer Biot during drying.
As a consequence of the above conclusions, numerical solutions
considering shrinkage and variable diffusion coefficient are sug-
gested for drying kinetic studies in high moisture foods.
Appendix.
The Halsey equations are as follows:
  1
Expð2:502Þ 0:9017
W¼ ðA1Þ
Lnaw
fitted with r2 = 0.996, and valid for 0.855 6 aw 6 0.999 while
  1
Expð2:367Þ 1:100
W¼ ðA2Þ
Lnaw
fitted with r2 = 0.998 and valid for 0.1 6 aw < 0.855.
References
Aguilera, J.M., 2005. Why food microstructure? Journal of Food Engineering 67, 3–
11.
Akpinar, E., Midilli, A., Bicer, Y., 2003. Single layer behaviour of potato slices in a
convective cyclone dryer and mathematical modelling. Energy Conversion and
Management 44, 1689–1705.
Adhikari, B., Howes, T., Lecomte, D., Bhandari, B.R., 2005. A glass transition
temperature approach for the prediction of the surface stickiness of a drying
droplet during spray drying. Powder Technology 149, 168–179.
Brooker, D.B., Bakker-Arkema, F.W., Hall, C.W., 1992. Drying and Storage of grains
and oilseeds. Van Nostrand Reinhold, New York.
Bruce, D.M., 1985. Exposed-layer barley drying: three models fitted to new data up
to 150 °C. Journal of Agricultural Engineering Research 32, 337–347.
Beuchat, L.R., 1981. Microbial stability as affected by water activity. Cereal Foods
World 26 (7), 345–349.
Beuchat, L.R., 1983. Influence of water activity on growth, metabolic activities and
survival of yeasts and molds. Journal of Food Protection 46, 135–141.
Chen, X.D., Peng, X., 2005. Modified Biot number in the context of air drying of small
moist porous objects. Drying Technology 23, 83–103.
Cicuttin, S.R., Giner, S.A., 2006. Geles pécticos semideshidratados dietéticos de
manzana. Simulación numérica considerando contracción volumétrica
(Semidehydrated low-calorie apple pectic gels. Numerical simulation
considering shrinkage). In Anais do Simposio Brasileiro Sobre
Desenvolvimento de Novos Produtos Alimenticios. Alimentos Funcionais e
para Fins Especias. Campinas-SP, Brazil.
Crank, J., 1975. The Mathematics of Diffusion, second ed. Oxford University Press,
Oxford, UK.
Crapiste, G.H., Rotstein, E., 1997. Design and performance evaluation of dryers. In:
Valentas, K.J., Rotstein, E., Singh, R.P. (Eds.), Handbook of Food Engineering
Practice. CRC Press, Boca Raton, pp. 125–165.
Gastón, A., Abalone, R., Giner, S.A., Bruce, D.M., 2004. Effect of modelling
assumptions on the effective water diffusivity in wheat. Biosystems
Engineering 88 (2), 175–185.
Gely, M.C., Giner, S.A., 2007. Diffusion coefficient relationships during drying of soya
bean cultivars. Biosystems Engineering 96 (2), 213–222.
Giner, S.A., Denisienia, E., 1996. Pressure drop through wheat as affected by air
velocity, moisture content and fines. Journal of Agricultural Engineering
Research 63, 73–86.
Giner, S.A., Mascheroni, R.H., 2001. Diffusive drying kinetics in wheat. Part 1:
potential for a simplified analytical solution. Journal of Agricultural Engineering
Research 80 (4), 351–364.
Giner, S.A., Mascheroni, R.H., 2002. Diffusive drying kinetics in wheat. Part 2.
Applying the simplified analytical solution to experimental data. Biosystems
Engineering 81 (1), 85–97.
Giner, S.A., 2009. Influence of internal and external resistances to mass transfer on
the constant drying rate period in high moisture foods. Biosystems Engineering
102, 90–94.
Greensmith, M., 1998. Practical Dehydration, second ed. Woodhead Publishing Ltd,
Cambridge, England.
Gurney, H.P., Lurie, J., 1923. Charts for estimating temperature distributions in
heating and cooling solid shapes. Industrial and Engineering Chemistry 15 (11),
1170–1172.
Kechaou, N., Roques, M.A., Lambert, J.F., 1987. Diffusion in shrinking media: the case
of drying of gels. In: Jowitt, R. (Ed.), Physical Properties of Foods, vol. 2. Elsevier
Applied Science, London, UK, pp. 55–60.
Leiva Díaz, E., Giannuzzi, L., Giner, S.A., 2009. Apple pectic gel produced by
dehydration. Food and Bioprocess Technology 2 (2), 194–207.
Kaymak-Ertekin, F., Gedik, A., 2004. Sorption isotherms and isosteric heat of
sorption for grapes, apricots, apples and potatoes. LWT Food Science and
Technology 37, 429–438.
Marinos-Kouris, D., Maroulis, Z.B., 1995. Transport properties in the drying of solids.
In: Mujumdar, A.S. (Ed.), Handbook of Industrial Drying, second ed., vol. 1.
Marcel Dekker, Inc., New York.
Mohsenin, N.M., 1980. Thermal Properties of Food and Agricultural Materials.
Gordon and Breach, London, UK.
Parti, M., 1993. Selection of mathematical models for drying grain in thin layers.
Journal of Agricultural Engineering Research 54, 339–352.
Singh, R.P., Heldman, D.R., 1993. Introduction to Food Engineering, second ed.
Academic Press, New York.
Sokhansanj, S., Bruce, D.M., 1987. A conduction model to predict grain temperatures
in grain drying simulation. Transactions of the ASAE 30 (4), 1181–1184.
Welti-Chanes, J., Mujica-Paz, H., Valdez-Fragoso, A., Leon-Cruz, R., 2003.
Fundamentals of mass transport. In: Welti-Chanes, J., Vélez-Ruiz, J., Barbosa-
Cánovas, G.V. (Eds.), Transport Phenomena in Food Processing. CRC Press LLC,
Boca Raton, USA.